RESUMO
INTRODUCTION: The ability to predict risk for poor outcomes in Crohn's disease [CD] would enable early treatment intensification. We aimed to identify children with CD with complications at baseline and throughout the study period who are at risk for surgery 2 years from diagnosis. METHODS: Newly diagnosed children with CD were enrolled into a prospective, multicentre inception cohort. Disease characteristics and serological markers were obtained at baseline and week 12 thereafter. Outcome data including disease activity, therapies, complications and need for surgery were collected until the end of 104 weeks. A chi-square automatic interaction detection [CHAID] algorithm was used to develop a prediction model for early surgery. RESULTS: Of 285 children enrolled, 31 [10.9%] required surgery within 2 years. Multivariate analysis identified stricturing disease at baseline (odds ratio [OR] 5.26, 95% confidence interval [CI] 2.02-13.67 [p = 0.001]), and Paediatric Crohn's Disease Activity Index [PCDAI] >10 at week 12 (OR 1.06, 95% CI 1.02-1.10 [p = 0.005]) as key predictors for early surgery. CHAID demonstrated that absence of strictures at diagnosis [7.6%], corticosteroid-free remission at week 12 [4.1%] and early immunomodulator therapy [0.8%] were associated with the lowest risk of surgery, while stricturing disease at diagnosis [27.1%, p < 0.001] or elevated PCDAI at week 12 [16.7%, p = 0.014] had an increased risk of surgery at follow-up. Anti-OmpC status further stratified high-risk patients. DISCUSSION: A risk algorithm using clinical and serological variables at diagnosis and week 12 can categorize patients into high- and low-risk groups from diagnosis.